If you run marketing at a startup, you have already felt it. The page that used to pull traffic still ranks, but the clicks are thinner. An answer engine summarized the topic, cited three sources, and the searcher never scrolled. The question has quietly shifted from “How do I rank?” to “How do I get cited?”
The short answer is less exotic than most playbooks make it sound. You do not win AI citations with a separate trick. You win them when your page is crawlable, easy to extract, clearly tied to a credible source, and better than commodity content, and when the rest of the web backs up what you say. Google’s own May 2026 guide to generative AI search says AI Overviews and AI Mode are rooted in core Search ranking and quality systems, use retrieval-augmented generation and query fan-out, and do not require llms.txt, special AI markup, or special schema.
What follows is the full version of that answer for founders and marketing leads: surface by surface, with current data, startup-ready tactics, and a clear line between what vendors have confirmed and what is still inference.
How do you get cited by AI search engines?
Citation visibility comes down to four things. Your content has to be (1) crawlable and indexable so the engine’s bot can reach and render it; (2) a direct answer to the question, structured so it is easy to extract; (3) backed by real expertise and original evidence rather than recycled summary; and (4) corroborated across the web by earned media, LinkedIn, YouTube, and other trustworthy references that repeat your claims. Do all four and you are eligible across every major engine. Miss one and you tend to get summarized without the link.
What AI search is, and why your clicks are falling
Classic SEO optimized for a ranked list of links. Answer engines optimize for a synthesized response that may quote, paraphrase, and link a handful of sources, then send fewer people onward. Two terms describe the work of getting into those responses: AEO (answer engine optimization) and GEO (generative engine optimization). They point at the same idea under different labels: earning visibility inside AI-generated answers, not only on the results page. For Google Search specifically, Google says those practices are still part of SEO.
The click pressure is real, though narrower than the doom headlines suggest. Ahrefs re-ran its CTR study on December 2025 data and published the update on February 4, 2026: AI Overviews correlated with a 58% lower average click-through rate for the top-ranking page when an Overview was present. A separate Graphite and Similarweb analysis summarized by Search Engine Land looked at more than 40,000 large U.S. sites and found organic search traffic down 2.5% year over year, not the 25% to 50% collapse often claimed. Both findings can hold at once. You lose low-friction informational clicks, but the larger picture is redistribution, not evaporation.
Scale is why this matters now. In Google’s I/O 2026 keynote transcript, Sundar Pichai said AI Overviews had passed 2.5 billion monthly active users and AI Mode had passed 1 billion. BrightEdge’s one-year tracking put AI Overviews on approximately 48% of tracked queries, up from 30% a year earlier. The surface where your buyers ask questions has changed. So the question worth your time is which sources these engines pick, and whether yours qualifies.
How each major engine chooses and cites sources
The engines are not interchangeable. They differ in what they crawl, how they retrieve, and where citations even appear. Treat them separately.
Google AI Overviews and AI Mode
Google’s position is the clearest in the market. AI Overviews and AI Mode pull from the Google Search index. They use retrieval-augmented generation and query fan-out, issuing several related searches to assemble a broader supporting set, and they carry no extra technical requirements beyond being indexed and snippet-eligible. The same fundamentals still apply: crawlability, internal linking, clear text, strong media, and current Business Profile or Merchant Center data where relevant. Google also reports AI features inside Search Console’s normal Web reporting, not a separate AI report.
One finding is worth sitting with. Ahrefs analyzed 863,000 SERPs and 4 million AI Overview URLs in March 2026 and found only about 38% of cited URLs also ranked in the top 10 for that exact query. Roughly 31% came from positions 11 to 100, and another 31% came from beyond the top 100. Query fan-out means Google increasingly sources from related sub-query results, not just the visible SERP. A page that ranks #14 for the headline term can still be cited if it best answers an adjacent question. The same study found 18.2% of non-ranking citations were YouTube URLs, with YouTube making up 5.6% of all cited URLs, so multimedia is a real citation surface here. In late May 2026, Google also brought Preferred Sources into AI Overviews and AI Mode, giving user preference signals a more visible role in which links stand out.
ChatGPT Search
OpenAI says ChatGPT Search gives timely answers with links, searches automatically when it judges the web would help, and also lets users trigger search manually. Responses can carry inline citations; when they do not, users can open a Sources panel.
The detail that matters most for publishers is the crawler. OpenAI’s publisher FAQ says public websites can appear in ChatGPT search and that pages should not block OAI-SearchBot if they want content included in summaries and snippets. It also notes that ChatGPT appends utm_source=chatgpt.com to referral URLs, so those visits can be tracked in analytics.
What OpenAI has not published is a detailed source-ranking model. The honest read stays modest: ChatGPT searches when current information would help, links sources, depends on OAI-SearchBot access for summaries and snippets, and uses a range of relevance and reliability factors that are not fully public.
Perplexity
Perplexity is the most citation-forward surface by design. It presents itself as an answer engine built around concise, sourced answers. The most actionable publisher detail is crawl access. Perplexity’s robots.txt guidance says PerplexityBot respects robots.txt and will not index full or partial text content from pages that disallow it. It also says PerplexityBot is for search-style indexing and is not used to train foundation models.
As with ChatGPT, there is no public formula for source weighting. The practical rule is simple: if you want eligibility, do not accidentally block the crawler that powers the surface.
Google Gemini
Gemini Apps are less transparent than Google Search, but Google’s help docs clarify the user experience. Gemini may show sources and related links, including public websites, uploaded files, or connected Workspace content. Not every response includes sources. If Gemini quotes a large amount of text from a webpage, Google says it shows a link to that page in the sources list, and web images are linked to their sources.
On the developer side, Grounding with Google Search lets Gemini generate search queries, process web results, and return grounding metadata with web sources and citations. One control is worth knowing about. Google-Extended governs whether content Google crawls may be used for future Gemini model training and for grounding in Gemini Apps and Vertex AI, but Google states it does not affect Google Search inclusion or ranking.
Microsoft Copilot
For information-seeking chats, Microsoft says Copilot Chat and Agents can use web search by generating a short query from the prompt and sending it to Bing. The sources button shows the query and sources used. Microsoft also documents that web search query citations are available in Microsoft 365 Copilot Chat, but not in the Copilot pane inside apps such as Word or PowerPoint. The practical read is that classic Bing discoverability, crawlability, and clarity still drive Copilot visibility, but you will not see citations everywhere.
What the latest data actually says
A grounding section before the tactics. These are the figures worth quoting, each dated. The patterns repeat across engines, but keep in mind that most of these studies are correlational, not causal.
On the signals that move citations, Semrush’s content optimization study from January 14, 2026 found the strongest positive correlations were clarity and summarization at +32.8%, E-E-A-T signals at +30.6%, Q&A format at +25.5%, section structure at +22.9%, and structured data at +21.6%. Promotional tone correlated negatively. Its technical SEO study of 5 million cited URLs, published January 5, 2026, found the schema types most common on cited pages were Organization, Article, and BreadcrumbList, with Google AI Mode citing higher-schema pages than ChatGPT Search.
On where AI looks beyond your own site, Muck Rack analyzed more than 25 million cited links from ChatGPT, Claude, and Gemini across 17 industries in May 2026 and found earned media accounted for 84% of all AI citations, with journalism at 27%. Semrush also analyzed 325,000 prompts across ChatGPT, Google AI Mode, and Perplexity; Social Media Today summarized the March 2026 findings, including LinkedIn’s role as one of the most-cited domains in AI answers.
On traffic and intent, Previsible analyzed 1.96 million LLM-driven sessions from November 2024 through November 2025 and found AI traffic was only 0.13% of total sessions on average, but concentrated on higher-intent pages such as industry, tools, blog, and pricing pages. Adobe’s retail data, covered by Search Engine Land, showed traffic from AI sources to U.S. retail sites grew 393% year over year in Q1 2026, and 39% of surveyed consumers had used AI for shopping. Read together: smaller volume, higher intent, fast growth in some verticals. Measure value, not just sessions.
How to earn citations across engines
Treat citation eligibility as a stack, not a single lever. Work it roughly in this order.
Fix crawler access and indexability first. This is unglamorous and frequently the actual blocker. Google AI features need your page indexed and snippet-eligible. ChatGPT search summaries need OAI-SearchBot access. PerplexityBot should not be blocked if you want Perplexity to index the page’s text. Before touching a word of copy, audit robots.txt, CDN and WAF rules, JavaScript rendering, and server logs to confirm the bots you want can reach and render the page.
Publish original information, not competent summaries. Google’s guide says unique, expert-led, non-commodity content beats hacks over time. Muck Rack and Semrush both show engines leaning on earned, editorial, and genuinely educational sources. For a startup, that means your citation assets are original benchmarks, first-party data, implementation lessons, honest technical comparisons, real pricing context, and customer-pattern insight, not another “10 tips” post.
Write for extraction. The pages that get quoted are easy to quote: a short direct answer up top, semantic headings, Q&A blocks, comparison tables, clear evaluation criteria, and chunked sections. Semrush’s data ties clarity, Q&A formatting, and section structure to higher citation rates. Competitive prompts often favor pages with explicit “Best for,” “Pros,” “Cons,” and “Pricing” structure. A cornerstone page should read almost like a well-organized reference.
Use structured data as support, not magic. Google is explicit that special schema is not required and warns against overfocusing on it. Even so, cited pages disproportionately carry Organization, Article, and Breadcrumb schema. Use schema to clarify entities, authorship, article type, breadcrumbs, and products. Do not sell it internally as a citation switch. One nuance: Google deprecated FAQ rich results in May 2026, so FAQ schema no longer earns the old visible FAQ SERP treatment, even though clear FAQ content can still help users and machine readability.
Strengthen entity consistency and off-site corroboration. The strongest cross-source pattern is that engines do not rely only on your site. Keep your company, product, founder, and category claims consistent across your website and schema, your LinkedIn company and executive profiles, YouTube, earned media, customer stories, and Business Profile or Merchant Center where relevant. Earned media dominates citations, and LinkedIn and YouTube are overrepresented. That makes off-site authority part of the core strategy now rather than a separate PR silo.
Refresh where the query is time-sensitive. Freshness matters most for comparative and fast-moving topics. Refresh cornerstone pages with a visible dateModified, current examples, updated screenshots, latest vendor docs, and dated evidence blocks. Heed the traffic data, though: refreshing old traffic plays without improving citation-worthiness will not recover queries that now end inside the answer engine.
What startup and B2B teams should do first
If you have limited hours this quarter, sequence the work like this.
First, take the fastest wins on existing pages: confirm AI crawler access, add an answer-first summary to your best-performing posts, insert comparison tables and Q&A blocks where they genuinely help, and tighten entity and author signals.
Next, create the net-new assets engines actually cite: original data, technical comparisons, and founder-led educational content on LinkedIn and YouTube, not just blog posts.
Last, build a measurement stack that tracks citations and prompts, AI referral traffic through source tags such as utm_source=chatgpt.com, and the downstream conversions those higher-intent visits produce. That last number is the one to optimize for.
For teams that want help with the technical and editorial side together, Triaza’s AI Search & Answer Engine Optimization service pairs naturally with traditional Search Engine Optimization. If you need the baseline SEO foundation first, start with SEO Fundamentals for Startups or the deeper SEO and AEO guide.
Questions marketers keep asking
Do I need llms.txt?
Not for Google Search visibility. Google says you do not need machine-readable AI text files, Markdown files, or special markup to appear in its generative search features. The confusing part is Lighthouse: Chrome’s newer Agentic Browsing audit does check llms.txt, but Google also says that category is experimental, does not use the normal 0-100 Lighthouse score, and marks a missing file as not applicable rather than a failure.
So the practical answer is: publish one if you want a clean agent-readiness signal and a simple map for tools that choose to read it, but do not treat it as a ranking factor or citation guarantee. If your site returns a server error for /llms.txt, fix that. If it returns a normal 404, that is not the same as failing Google’s search guidance.
Do backlinks still matter?
Yes, reframed. Off-site corroboration through earned media, LinkedIn, YouTube, and credible references is one of the strongest cross-engine patterns. Muck Rack’s 2026 dataset found earned media made up 84% of AI citations. Think authority and consistency across the web, not raw link volume.
Why is my top-ranking page not cited?
Because citation overlap with the top 10 is weaker than people assume, around 38% in Ahrefs’ March 2026 data. Query fan-out pulls from related sub-queries, so the fix is usually to answer adjacent questions clearly and make the page easier to extract, not only to chase the headline rank.
Is AI search replacing SEO?
No. Every official source still points back to SEO fundamentals: crawlability, indexability, clarity, authority, and useful content. AEO extends that work; it does not replace it.
How do I measure AI visibility?
Combine three things: citation and prompt tracking, AI referral traffic in analytics, and conversions from those sessions. Do not judge the channel on session volume alone. It is low-volume and high-intent by nature.
Where to start this week
The fastest move is also the least glamorous: open your logs and robots.txt and confirm Googlebot, OAI-SearchBot, and PerplexityBot can actually reach your best pages. Most teams discover at least one block they did not know about. After that, pick your three highest-intent pages, add a direct answer at the top of each, and check whether your company and founder details say the same thing on your site, LinkedIn, and anywhere the press has quoted you. None of that requires a new tool or a bigger budget, and it is where citation eligibility is won or lost.
